Shooting on the run

[gnirts]

I know that our team is definitely going to try to make goals while the robot is moving in any given direction.

How many other teams think this is viable?

[X-Istence]

Originally Posted by gnirts

I know that our team is definitely going to try to make goals while the robot is moving in any given direction.

How many other teams think this is viable?

Depends on how the shooter is mounted and or how the robot is built. I think it is viable to shoot on the run, however, not for the design my team and I have come up with.

[gnirts]

I find it hard to believe that all teams can shoot at the center goal (at least at our regional I expect many teams, especially rookies, to go for the two lower goals).
... this is an anonymous poll!

[aaeamdar]

Originally Posted by gnirts

I find it hard to believe that all teams can shoot at the center goal (at least at our regional I expect many teams, especially rookies, to go for the two lower goals).
... this is an anonymous poll!

There are only 22 results so far (a small sample size) and I think your results are getting biased by the fact that you made your title suggest that it was only a thread for shooting teams. Our team, for the record, can't shoot in the center.

Also, I'm curious why this went in the programming forum, or did I misread that?

Paul Dennis

[gnirts]

Originally Posted by aaeamdar

... I think your results are getting biased by the fact that you made your title suggest that it was only a thread for shooting teams.

Oops, my bad. I am not sure how to change the title.

Also, I'm curious why this went in the programming forum, or did I misread that?

Well, I figured two things would make that a logical choice:

1. If anyone on the team knows how reliably they will be shooting on the run, its the programmers.
2. And of all the people that visit Chief Delphi, programmers are probably on it the most. (just a guess, though)

[paulcd2000]

Originally Posted by gnirts

And of all the people that visit Chief Delphi, programmers are probably on it the most. (just a guess, though)

definetly true

[greencactus3]

we defintly will be able to. but accuracy wise... i dunno if we'll actually do it in a match.

[Donut]

I doubt we'll do this (at least using camera control, our drivers can try it if they REALLY want to, but I wouldn't advise it). We don't have an accelermotor, and if we don't use the encoders either, trying something like this would be near impossible.

[gnirts]

Originally Posted by Donut

We don't have an accelermotor, and if we don't use the encoders either, trying something like this would be near impossible.

If you are talking about firing while moving, all you need to know is your speed, which way your turret is aimed in relation to the direction of the robot, which way the robot is pointed, and have tracking data on the green light. Maybe not even that much.

[Rick TYler]

Originally Posted by gnirts

If you are talking about firing while moving, all you need to know is your speed, which way your turret is aimed in relation to the direction of the robot, which way the robot is pointed, and have tracking data on the green light. Maybe not even that much.

To build an effective anti-ballistic missile system, all you need to know is the starting point of the incoming missile, its boost, its average speed, its parabolic trajectory and its evasive capabilities. It's a simple problem...

[gnirts]

Originally Posted by Rick TYler

To build an effective anti-ballistic missile system, all you need to know is the starting point of the incoming missile, its boost, its average speed, its parabolic trajectory and its evasive capabilities. It's a simple problem...

Reminds of in the beginning of the season when the strategy team wanted me to use a combination of this year's camera, last year's camera, and an ultrasonic range finder to knock other teams balls out of the air

But on a more serious note, would you not also need to know where you are in relation to said projectile? And in FIRST, is knowing your exact position on the field less than trivial (esp. with non Kevin-ized accelerometer code)?

Good luck to all,
Robinson

[Donut]

Originally Posted by gnirts

If you are talking about firing while moving, all you need to know is your speed, which way your turret is aimed in relation to the direction of the robot, which way the robot is pointed, and have tracking data on the green light. Maybe not even that much.

To know speed, you must have an accleremotor or an encoder, as I said. Robot turning requires gyro, angle of rotator in relation to the robot requires a POT, and the tracking stuff requires the camera. I hope all teams will have the camera at least semi-working, but using all the other data is alot to ask for.

You must know your speed in your current direction, relate this to how fast you are moving to/from the goal, and change your shooter speed/rotation to account for it. Is your shooter or rotator going to be able to move/change speed fast enough to keep up? By the time it gets in position, your robot might be well past the point that required those parameters. Also, what happens if your robot is not moving at a constant speed? You must now make a part of your code to account for acceleration in addition to velocity, even more of a pain. As an added bonus, a strong hit from another robot always has the possibility of throwing off your data (you may think parallel to the goal is 5 degrees past the original angle you thought of), compromising the whole system.

All of these calculations and compensations are possible. I don't think there is enough time in 6 weeks to do them all and test them properly though. I say less than 5 teams are able to do this well, and at least 4 of them will be from teams that used the camera very effectively in autonomous last year.

[gnirts]

Originally Posted by Donut

To know speed, you must have an accleremotor or an encoder, as I said. Robot turning requires gyro, angle of rotator in relation to the robot requires a POT, and the tracking stuff requires the camera. I hope all teams will have the camera at least semi-working, but using all the other data is alot to ask for.

You must know your speed in your current direction, relate this to how fast you are moving to/from the goal, and change your shooter speed/rotation to account for it. Is your shooter or rotator going to be able to move/change speed fast enough to keep up? By the time it gets in position, your robot might be well past the point that required those parameters. Also, what happens if your robot is not moving at a constant speed? You must now make a part of your code to account for acceleration in addition to velocity, even more of a pain. As an added bonus, a strong hit from another robot always has the possibility of throwing off your data (you may think parallel to the goal is 5 degrees past the original angle you thought of), compromising the whole system.

All of these calculations and compensations are possible. I don't think there is enough time in 6 weeks to do them all and test them properly though. I say less than 5 teams are able to do this well, and at least 4 of them will be from teams that used the camera very effectively in autonomous last year.

Knowing your current vector is important, but setting up encoders or the GTS is not hard. You could also consider having a lookup table of PWM values to approx.. speed (of course, this would not be very accurate, but it might work with some tweaking). You could also see how many pixels the target moves each frame and use this in combination with your distance from the target to deduce an approx. speed. A tachometer could be used as an analog input to get speed as well.

You need to also know the current direction, but the gyro is so easy to setup this couldn't be a problem. Acceleration is, in my mind, irrelevant, because the ball does not retain the acceleration of the robot once it has left the turret, just the velocity of the robot at the point it left. And being bumped by a robot won't occur continuously, nevertheless it will be a factor that is impossible to account for.

[If you need help with either encoders, gyro, or the GTS there are plenty of help posts and Kevin's code is well documented and clear]

However, your point about never being able to move the turret accurately and fast enough to have it in the position to fire after doing the calculations is my worst fire. I can't test any of this stuff because our robot is ages away from being ready. In my mind, its the the most difficult problem to overcome. Of course, any one who has mode can feel free to give us some hints.

Good Luck,
Robinson

[aaeamdar]

Originally Posted by gnirts

1. If anyone on the team knows how reliably they will be shooting on the run, its the programmers.
2. And of all the people that visit Chief Delphi, programmers are probably on it the most. (just a guess, though)

It's not that I mind that you posted it here, and it seems like you had a fine rationale. I was just curious as to the reason, thanks for the reply =).